Matthias Hinrichs
54 lines
1.8 KiB
Python
54 lines
1.8 KiB
Python
"""Phonemization with espeak-ng."""
|
|
|
|
import re
|
|
import unicodedata
|
|
from pathlib import Path
|
|
from typing import Union
|
|
|
|
_DIR = Path(__file__).parent
|
|
ESPEAK_DATA_DIR = _DIR / "espeak-ng-data"
|
|
|
|
|
|
class EspeakPhonemizer:
|
|
"""Phonemizer that uses espeak-ng."""
|
|
|
|
def __init__(self, espeak_data_dir: Union[str, Path] = ESPEAK_DATA_DIR) -> None:
|
|
"""Initialize phonemizer."""
|
|
from . import espeakbridge # avoid circular import
|
|
|
|
espeakbridge.initialize(str(espeak_data_dir))
|
|
|
|
def phonemize(self, voice: str, text: str) -> list[list[str]]:
|
|
"""Text to phonemes grouped by sentence."""
|
|
from . import espeakbridge # avoid circular import
|
|
|
|
espeakbridge.set_voice(voice)
|
|
|
|
all_phonemes: list[list[str]] = []
|
|
sentence_phonemes: list[str] = []
|
|
|
|
clause_phonemes = espeakbridge.get_phonemes(text)
|
|
for phonemes_str, terminator_str, end_of_sentence in clause_phonemes:
|
|
# Filter out (lang) switch (flags).
|
|
# These surround words from languages other than the current voice.
|
|
phonemes_str = re.sub(r"\([^)]+\)", "", phonemes_str)
|
|
|
|
# Keep punctuation even though it's not technically a phoneme
|
|
phonemes_str += terminator_str
|
|
if terminator_str in (",", ":", ";"):
|
|
# Not a sentence boundary
|
|
phonemes_str += " "
|
|
|
|
# Decompose phonemes into UTF-8 codepoints.
|
|
# This separates accent characters into separate "phonemes".
|
|
sentence_phonemes.extend(list(unicodedata.normalize("NFD", phonemes_str)))
|
|
|
|
if end_of_sentence:
|
|
all_phonemes.append(sentence_phonemes)
|
|
sentence_phonemes = []
|
|
|
|
if sentence_phonemes:
|
|
all_phonemes.append(sentence_phonemes)
|
|
|
|
return all_phonemes
|